Your assessment is dead on. This is becoming more common in research related positions. For many years, I've "paid" almost the entirety of my salary in similar roles, I essentially pay a toll to an institution for its name, prestige, and connections as well as networking opportunities to support my continued employment.
The reason you don't set starting salaries too high is that for the business managing this (be it a university, private lab, etc.), although they essentially hire someone who makes money for them and pays their salaries through overhead costs that almost always exist (indirect costs), if you set the salary too high the role won't be able to reliably maintain itself, and therefor the indirect percentage you draw from incoming grants from that role becomes less stable. You could set a research associate for a fresh post doc at say $500k, but they'll probably never be able to pay themselves and therefor you because they won't be able to secure that. So you'll constantly be hiring for that position so it has to drop down to an amount that can be reliably filled and reliably bring in money.
From the institution's perspective, they want to get your comp rate as high as they possibly can to maintain their salaries but not to a point where it self sabotages their own position. The way these institutions maintain their administrative positions is to find these sweet spots and then diversify risk by hiring more and more researchers to pull in money for themselves and their own salaries. If several fail to pull in grants, admin is cushioned by the total number of grants (likely enough pulled in grants to provide enough indirect for admin). Take the additional pool and reinvest it in research so you get researchers to do even more free work for you. Some research orgs I've seen have incredibly high turn over rates (on the order of 1-2 years) and their model seems to be as long as they can keep a fresh batch of unsuspecting individuals in at high enough rates they can draw grants and indirect from to pay their salaries, it doesn't matter if that individual hired becomes resource pressured and burns out, there's a line of other qualified people more than happy to step in and naively fall into the same game. Rinse repeat.
Eventually it can tarnish the reputation of the organization itself where external money pools stop looking at the person receiving the grant but more at the organization of the person receiving the grant where they begin to lose faith that the organization can follow through on its employees promises. There seems to be enough turnover from funding agencies and the research orgs that do this themselves that institutional memory is short enough to keep the whole game going on. It's getting to a point at some organizations where people who pull in grants are joining the same gig as the research org and the idea is to act as a figurehead to the project where they then convince naive other lower level research positions to do most the legwork and pay themselves a portion of the total grant acting as another layer of administration. You'll see such grant recipients often as the recipient or lead on multiple grants they couldn't possibly be working on all of simultaneously. They play similar risk games to the research orgs that employ them in that if those they hire can't follow through, they may lose their own role.
It's quite a gig in some places, a bit of a pyramid scheme, really. This is why I've lost so much faith in research publications anymore, because the underbelly is pretty ugly. I've seen this at universities, private labs, even national labs and "think tanks." We really need to reevaluate and restructure how research is performed in the US.